Edge based location feature index matching

1. A method for image processing, said method comprising: (a) a computing device, labeling edge pixels of a model image using an edge based technique; (b) said computing device, specifying an angular orientation for each of said edge pixels of said model image, wherein a plurality of pixels of said model image are labeled as non-edge pixels and do not have an angular orientation specified; (c) said computing device, labeling edge pixels of an input image using an edge based technique; (d) said computing device, specifying an angular orientation for each of said edge pixels of said input image; (e) said computing device, indexing each of said angular orientations for each of said edge pixels of said input image into a feature index identifying its respective location wherein said respective location is stored as a one-dimensional value in a location list, where the one-dimensional value preserves a two-dimensional said location; (f) said computing device, determining candidate locations of an object within said input image based on said feature index and said angular orientation for each of said edge pixels of said model image, and including identifying said candidate locations in a coarse resolution stage that identifies for at least one block of pixels a predetermined number of top-ranked said angular orientations, ranking made by a criteria that includes at least one of: (i) the frequency of the occurrence of a said angular orientation in a block of pixels; (ii) the strength of an edge associated with a said angular orientation; and (iii) a gradient magnitude of an edge associated with a said angular orientation where the top-ranked said angular orientations are used to define a low resolution template, and where the candidate locations are modified in a fine resolution stage using at least one high resolution template; and (g) based upon the modified said candidate locations selecting at least one object location within said input image.

2. The method of claim 1 wherein a plurality of pixels of said input image are labeled as non-edge pixels and do not have an angular orientation specified.

3. The method of claim 1 wherein said angular orientations of at least one of said model image and said input image are defined relative to a reference angle as, +90 degrees to +60 degrees, +60 degrees to +30 degrees, +30 degrees to +0 degrees, +0 degrees to −30 degrees, −30 degrees to −60 degrees, −60 degrees to −90 degrees, −90 degrees to −120 degrees, −120 degrees to −150 degrees, −150 degrees to −180 degrees, +180 degrees to +150 degrees, +150 degrees to +120 degrees, and +120 degrees to +90 degrees.

4. The method of claim 3 wherein said reference angle is 15 degrees with respect to a horizontal axis.

5. The method of claim 1 wherein said angular orientations of at least one of said model image and said input image are represented relative to a reference angle as, +90 degrees to +60 degrees as 00000001, +60 degrees to +30 degrees as 00000010, +30 degrees to +0 degrees as 00000100, +0 degrees to −30 degrees as 00001000, −30 degrees to −60 degrees as 00010000, −60 degrees to −90 degrees as 00100000, +90 degrees to +60 degrees as 00000001, +60 degrees to +30 degrees as 00000010, +30 degrees to +0 degrees as 00000100, +0 degrees to −30 degrees as 00001000, −30 degrees to −60 degrees as 00010000, 60 degrees to −90 degrees as 00100000, −90 degrees to −120 degrees as 00000001, −120 degrees to −150 degrees as 00000010, −150 degrees to −180 degrees as 00000100, +180 degrees to +150 degrees as 00001000, +150 degrees to +120 degrees as 00010000, and +120 degrees to +90 degrees as 00100000, −90 degrees to −120 degrees as 00000001, −120 degrees to −150 degrees as 00000010, −150 degrees to −180 degrees as 00000100, +180 degrees to +150 degrees as 00001000, +150 degrees to +120 degrees as 00010000, and +120 degrees to +90 degrees as 00100000.

6. The method of claim 5 wherein said reference angle is 15 degrees with respect to a horizontal axis.

7. The method of claim 1 wherein said angular orientations of at least one of said model image and said input image are represented relative to a reference angle as at least four different angular orientations, each of which substantially span the same angular range.

8. The method of claim 1 wherein said angular orientations of at least one of said model image and said input image are represented relative to a reference angle as, +75 degrees to +45 degrees as 00000011, +45 degrees to +15 degrees as 00000110, +15 degrees to 45 degrees as 00001100, 45 degrees to −45 degrees as 00011000, −45 degrees to −75 degrees as 00110000, −75 degrees to −105 degrees as 00100001, −105 degrees to −135 degrees as 00000011, −135 degrees to −165 degrees as 00000110, −165 degrees to +165 degrees as 00001100, +165 degrees to +135 degrees as 00011000, +135 degrees to +105 degrees as 00110000, and +150 degrees to +75 degrees as 00100001.

9. The method of claim 5 wherein said angular orientations of said at least one of said model image and said input image are further represented relative to said reference angle as, +75 degrees to +45 degrees as 00000011, +45 degrees to +15 degrees as 00000110, +15 degrees to −15 degrees as 00001100, −15 degrees to −45 degrees as 00011000, −45 degrees to −75 degrees as 00110000, −75 degrees to −105 degrees as 00100001, −105 degrees to −135 degrees as 00000011, −135 degrees to −165 degrees as 00000110, −165 degrees to +165 degrees as 00001100, +165 degrees to +135 degrees as 00011000, +135 degrees to +105 degrees as 00110000, and +150 degrees to +75 degrees as 00100001.

10. The method of claim 1 wherein determining candidate locations of an object within said input image is based upon spatial characteristics of said model image.

11. The method of claim 1 wherein at least one of said angular orientations for said edge pixels of said input image map to a plurality of said feature indexes.

12. The method of claim 1 wherein at least one of said angular orientations for said edge pixels of said model image map to a plurality of angular orientations for said matching.

13. The method of claim 1 wherein at least one of said angular orientations for said edge pixels of said input image map to a plurality of said feature indexes and at least one of said angular orientations for said edge pixels of said model image map to a plurality of angular orientations for said matching.

14. A method for image processing comprising: (a) a computing device determining edge pixels of a first model template and a second model template using an edge based technique, said first model template and said second model template being rotated versions of a single model image; (b) said computing device determining an angular orientation for each of said edge pixels of said first model template and said second model template; (c) said computing device determining edge pixels of an input image using an edge based technique; (d) said computing device determining an angular orientation for each of said edge pixels of said input image; (e) said computing device indexing each of said angular orientations for each of said edge pixels of said input image into a feature index identifying its respective location; (f) said computing device determining candidate locations of an object within said input image based on said feature index and said angular orientation for each of said edge pixels of said first model template and said second model template, using a single score map; (g) based upon said candidate locations selecting at least one object location within said input image.